At fixed periods, HD 20794 circular-pair trials give tail rates of 0.9% for white noise and 17.9% for 30-day noise

We test two explanations for the public HD 20794 velocities: one eccentric orbit, or two circular orbits. The outer period is fixed at 647.6 days. The circular pair has a second period of 323.8 days. We use 811 night means and three fixed Gaussian noise matrices. Each model has 11 nominal mean parameters. We generate 1,000 trials from each fitted model mean and fit both complete model families again. Two nearby circular-pair means provide additional tests in each noise setting. For independent noise, the observed fit advantage is matched or exceeded by 9 of 1,000 circular-pair trials. The rate is 0.9 percent, with a 95 percent Monte Carlo interval of 0.41 to 1.70 percent. The 30-day setting uses a 0.5 m/s exponential noise term and the same total marginal variance. Its rate is 179 of 1,000 trials, or 17.9 percent (15.57 to 20.42 percent). A 300-day control gives 0.6 percent (0.22 to 1.30 percent). Its circular-pair fit also has a poor absolute chi-square upper tail of 0.00127. Both correlated settings place the observed eccentricity at the imposed ceiling of 0.8. The relative preference therefore depends on the tested noise assumption. These are local trial rates at fitted means. They are not uniform p values for all circular-pair parameters or planet probabilities. No new planet is detected. The package contains source tables, provenance, code, all 12,000 trials, figures, and numerical checks.

HD 20794 · radial velocity · eccentricity · circular orbit pair · parametric simulation · correlated noise · model comparison · reproducible astronomy